Okun’s law and youth unemployment in Germany and Poland

Sophie Dunsch

___________________________________________________________________

European University Viadrina Frankfurt (Oder)

Department of Business Administration and Economics

Discussion Paper No. 373

September 2015

ISSN 1860 0921

___________________________________________________________________

Okun’s law and youth unemployment in Germanyand Poland

Sophie Dunsch∗

Abstract

The unemployment rates, especially youth unemployment rates, in-creased in various countries of Europe over the last years. This paper ex-amines youth unemployment developments in Germany and Poland withOkun’s law to test the hypothesis that young employees are more exposedto the business cycle. I estimate age and country specific Okun coeffi-cients for five different age cohorts. The results show that youth in Polandis more sensitive to the business cycle than adults, while in Germany thedifference between the age cohorts is not that distinctive.A further examination of the different labor market institutions regard-ing youth employment results in policy recommendations beyond GDPgrowth, such as job-search assistance as short-term and structural reformsregarding education as long-term recommendation.

Keywords: Youth Unemployment, Okun’s Law, Poland, Germany

JEL classification: E24, J64

∗Faculty of Business and Economics, European University Viadrina, Chair of Macroeco-nomics, Große Scharrnstrasse 59, Frankfurt (Oder), 15230, Germany, E-mail: dunsch@europa-uni.de

1

1 Introduction

The financial and economic crisis strongly affected the European labor mar-kets, but with different outcomes in the different countries. I investigate theunemployment development in Germany and Poland, because their cases in therecession are special. In Germany, the youth unemployment rate had been quitestable after the financial crisis, even declining after 2009. But the developmentof the growth rate of the real gross domestic product (GDP) was as expected,i.e. there was negative GDP growth in 2009. In contrast, Poland had permanentpositive GDP growth rates, but the youth unemployment rate increased. EU-15 countries as an aggregate, which includes all countries that were membersof the European Union before the eastern enlargement in May 2004, is usedfor comparison. Using Okun’s law (Okun, 1962), which expresses a negativerelationship between changes of the unemployment rate and the growth rateof the GDP, I examine whether youth is more sensitive to the business cyclethan adults (Boulhol and Sicari, 2013). My hypothesis is that if the economyis in a recession, young employees are the first to be dismissed and thereforemore vulnerable to cyclical shocks. Additionally, I examine how strong thedifferences between the various age cohorts are and therefore estimate age andcountry specific Okun coefficients for five age cohorts. The results show thatyouth in Poland is more prone to the business cycle conditions than adults,while in Germany the difference between the age cohorts is not that distinctive.This result will then lead to an examination of the two labor markets to findthe causes of those differences. As there are labor market institutions whichaffect youth unemployment more than adult, I am examining e.g. the degree ofemployment protection legislation for different types of contracts, the minimumwage and the extent to which temporary contracts are used (Berlingieri et al.,2014, Brada et al., 2014). Policy recommendations, beyond the need of GDPgrowth, will be tackling the demand as well as the supply side of the labormarket.The structure of the paper is as follows: Section 2 provides a short literatureoverview regarding the main aspects of youth unemployment. Section 3 de-scribes the data set. In Section 4 I discuss the empirical results according toOkun’s law. Section 5 examines the labor market institutions of Germany andPoland to explain the differences found in Section 4, while Section 6 concludesthe paper and recommends a course of action, divided in short- and long-termproposals.

2

2 Literature review

The link between unemployment and real GDP growth can be explained fromthe demand side. An increase in aggregated demand will lead to an increase inproduction. This will lead to an increase in demand for labor and therefore toa decline of the unemployment rate. Following this line of reasoning, a negativeshock in the GDP will lead to a lower demand for labor and therefore to a risein the unemployment rate. This is valid for the whole labor market as well asfor different age cohorts (O’Higgins, 1997).The unemployment rate depends on various country-specific factors, e.g. theextent of „skills mismatch“ and the transition from school to work are influ-encing the level of the youth unemployment rate (Dietrich, 2012). However,changes in the youth unemployment rate can also be caused by cyclical fluc-tuations. Young people are more sensitive to cyclical changes, because thecompanies have lower opportunity costs when discharging young employees.Following O’Higgins (1997) this is due to the fact that young employees haveless company-specific skills and less dismissal protection in comparison to olderemployees. In addition, Bell and Blanchflower (2011) argue that youth findsitself in a so-called „experience trap“, i.e. employers select workers with ex-perience, and as a result, labor market entrants are never hired and thereforecannot increase their own experience. This might lead to higher unemploymentrates for young people especially in an economic downturn where they mustcompete with more experienced and skilled adults for fewer jobs (Unt, 2012).In contrast, it is argued that youth unemployment is of shorter duration andless problematic, because young people would change their workplace easily andmore often and look for a more appropriate (skill-matching) position (O’Higgins,2003). But even if youth is experiencing a shorter unemployment duration, itcan have other effects: Berlingieri et al. (2014) argue that a failure in integrat-ing the young generation implies a loss of production, productivity and verylikely also a loss in innovation potential. Furthermore, there is a fiscal cost ofyouth unemployment due to increased welfare payments and loss of tax rev-enues besides the associated loss of GDP (Berlingieri et al., 2014).Additionally, Mroz and Savage (2006) find that unemployment in young yearshas profound negative effects on human capital accumulation leading to lowerearnings in the future. Youth unemployment today will lead to higher socialcost in the future and negative impacts on wellbeing, health status and jobsatisfaction (Bell and Blanchflower, 2011). Further effects can be deskilling anda degradation of physical and mental health (Berlingieri et al., 2014).With this in mind, I would like to have a closer look at the hypothesis that if

3

the economy is in an economic downturn, young employees are the first to bedismissed and therefore more sensible to cyclical shocks.

3 Data set and descriptive statistics

The data set consists of annual real GDP, measured in prices of the year 2010and published in the Annual Macro-Economic Database (AMECO) of the Eu-ropean Commission (EC, 2015). The annual unemployment rates for variousage cohorts are provided by the Organisation for Economic Co-operation andDevelopment (OECD, 2015b). It uses the earliest available entries for eachcountry (Germany: 1992, Poland: 1993 and EU-15: 1992) and ends in 2014.The unemployment rate is based on International Labour Organisation (ILO)standards to ensure the comparability among the countries.Figure 1 shows GDP growth in Germany, Poland and EU-15. Poland has onlypositive GDP growth rates during the financial crisis, while Germany and EU-15show a negative GDP growth in 2009. Figure 2 highlights the youth unemploy-ment rates for Germany, Poland and EU-15 from 1992 until 2014, i.e. for theage cohort of the 15-to-24-years old. The rates vary between the countries: Ger-many has very low rates and even after the crisis, those rates decline. Polandhad declining rates before the crisis, but after 2009 the rates rose again despitethe fact that Poland had always positive GDP growth, even during the crisis.The EU-15 as aggregate had the expected increase in unemployment rates afterthe financial crisis. For all countries the rates are decreasing in 2014.The youth-to-adult unemployment ratio shows if the employment prospectsof youths are worse than those of adults who participate in the labor market(Berlingieri et al., 2014, Bell and Blanchflower, 2011). The ratio is calculatedby dividing the youth (15-to-24-year-olds) unemployment rate by the adult (25-to-64-year-olds) unemployment rate. It measures whether youth or adults areexperiencing larger difficulties in the labor market and a higher ratio indicatesthat youth suffers disproportionately to adults. Figure 3 shows the ratios cal-culated for Germany, Poland and EU-15. In Poland, the youth unemploymentrates are more than twice than those of adults, while in Germany, the ratio isconsiderable smaller. But in Germany as well as in Poland, even before thegreat recession this ratio increases, while in EU-15, the adult unemploymentrates increases more rapidly, showing a slightly decreasing youth unemploy-ment rate/adult unemployment rate ratio after 2008.In the next section I am examining in detail the relationship between youthand adult unemployment rate.

4

4 Regression analysis

4.1 Relationship between youth and adult unemployment rates

I analyze the relationship between youth and adult unemployment rates by re-gressing the youth unemployment rate on the adult rate (Bell and Blanchflower,2011, O’Higgins, 2012). The equation can be written as:

uyit = αi + γiuait + �it, (1)

where uyit is the youth unemployment rate (age cohort 15-24) for country iat time t and uait is the corresponding adult unemployment rate (age cohort25-64). This simple analysis does not consider other factors, such as cohortsize, prices or marginal products of youth and adult labor. The results areshown in Table 1 for Germany, Poland and EU-15.The unemployment rate of the youngest age cohort in Poland changes by 2.19%for each 1% change in adult rates. While in the EU-15 it is as high as in Polandwith 2.18%, in Germany the youth unemployment rate changes by 1.01%for a change of 1% in the unemployment rate of adults. The German resultcould be interpreted as if youth and adults are complements and a decreasein adult unemployment is accompanied by a decrease in youth unemployment(O’Higgins, 2012).This result confirms my hypothesis that young employees, especially in Poland,are more sensitive to changes in aggregate demand of labor as adults (OECD,2009). In the next step I would like to answer the question: how strong is thedifference between the different age cohorts?

4.2 Okun’s law

There are several versions of Okuns’ law. The original ones were proposed byOkun (1962), the so-called gap and difference version. Furthermore, there arederivations developed over time, so-called dynamic versions (see, for example,Knotek, 2007). Here, the difference version will be used to analyze the sensi-tivity of the unemployment rate to changes in the growth rate of GDP. Theregression is given by:

∆uit = αi + βiGDPgrowthit + �it, (2)

5

where ∆uit is the change in the unemployment rate from period t − 1 to t forcountry i, GDPgrowthit represents the GDP growth rate1 and �it is an assumedwhite noise error term. The parameter βi is the so-called „Okun coefficient“.According to Okun’s law, the coefficient should be negative, i.e. positive GDPgrowth should lead to a decrease of the unemployment rate (Hutengs and Stadt-mann, 2014b).In addition to the regression via Ordinary Least Squares (OLS)2, a balancedpanel for each country is constructed and used for further estimations. Thispanel resolves the problem that there is only a limited number of observationsavailable for single OLS estimates. The panel includes the yearly changes inthe unemployment rate and the GDP growth rate for five different age cohorts.Rather than estimating each beta coefficient for each age cohort separately, thepanel will be estimated via a least squares dummy variable model (LSDV) foreach country:

∆uj,t = αjDj + βjDjGDPgrowtht + �j,t, (3)

where ∆uj,t is the change in unemployment rate for cohort j at time t, Djsymbolizes a dummy variable representing the different age cohorts and �j,t isan assumed white noise error term. The parameters βj capture the differentcohort specific Okun coefficients.The OLS residuals have been checked for heteroscedasticity and serial corre-lation and I found both in all country panels (see test results in Table 2 andTable 3). As heteroskedasticity and autocorrelation may lead to inefficientestimates with biased standard errors and thus misleading results, I fitted themodel with MA(1) errors. The results are shown in Table 4.The Okun coefficients are negative across all countries and age cohorts. Thus,the expected negative relationship between changes in the unemployment rateand the real GDP growth can be confirmed. The strength of the effect differsbetween all countries which is expected due to different labor markets. Allcountries as well as the aggregate EU-15 show their highest absolute Okuncoefficients among the age cohort of the 15-to-24 years old. This indicates thatyoung people are more sensitive to the business cycle conditions than adults,especially in comparison to the age cohort of the 55-to-64-years old.There are differences between the countries. In Poland, Okun’s coefficientsin absolute values are larger than in Germany, so the Polish youth suffers1The GDP growth rate has been calculated as a percentage change in GDP moving fromGDPt−1 to GDPt: GDP growtht =

(GDPt−GDPt−1

GDPt−1

)· 100.

2Estimation results can be requested from the corresponding author.

6

disproportionately more than the German youth. This supports the resultregarding the relationship between youth and adult unemployment rates insection 4.1.In Poland, the strongest increase in the Okun coefficient is observed from the15-24 years cohort to the 25-34 years cohort, while in Germany the differencesare not that distinct. In Germany, the strongest increase is observed betweenthe 25-34 years cohort and the 35-44 years cohort. This might be the case,because the younger age cohort includes the ones finishing tertiary educationand searching for the first job, while in comparison, the older age cohort as wellas the age cohort of the 45-54 years old are mostly well established on the labormarket due to their working experience as well as their accumulated skills.Bell and Blanchflower (2011) argue that young people may be less efficientin job search activities than adults and are likely to have fewer contacts andless experience of finding work which places them at a relative disadvantagecompared to adults.A closer look at the age cohort of the 25-34 years old is appropriate, as theyalready completed their education and enter the labor market the first time(Pastore, 2015). In Table 8 the unemployment as well as the labor marketparticipation rates of the 25-34 years old are shown in comparison to the ratesof the 15-24 years old. The unemployment rates are lower, even when theirlabor market participation rate is higher. The youngest age cohort is still theone that should be in the focus of the analysis.In all countries as well as in the EU-15 countries as aggregate, the smallestOkun coefficients in absolute values are the ones for the 55-to-64 years old.This could be a result of better protection by employment protection laws.Therefore, this age cohort is less vulnerable to business cycles, because theyare the last to lose their job in a recession (Hutengs and Stadtmann, 2014b).In addition, the equality of coefficients for each country between age cohortshas been tested with a Wald-Test and the results are shown in Tables 5 to 7.3

Only for EU-15 the test confirms that the coefficients for youth (age 15-24)differs significantly from those of older age cohorts, while in Poland there isonly a significance for the differences to two oldest age cohorts.Business cycle effects are not explaining all country differences in the levelof youth unemployment. The youth-to-adult unemployment ratio, which hasbeen shown in section 3 and can be seen as an indicator of potentially existingstructural problems (Cahuc et al., 2013), points at differences between thecountries. Therefore, in the next section I examine the labor markets Poland3Each table show empirical F-Values and the corresponding significance level for one country.

7

and Germany in detail to answer the question what the underlying causes forthe differences between the age cohorts and countries could be.

5 Labor markets in detail

There are different aspects in the labor market that may affect youth employ-ment and explain the differences between the age cohorts as found in section 4.Those aspects include e.g. the degree of employment protection legislation fordifferent types of contracts, etc. (Berlingieri et al. (2014)). According to Bradaet al. (2014), besides institutional variables such as labor taxes, unemploy-ment benefits, unionization and collective bargaining, some specific variablesfor youth unemployment include the minimum wage and the extent to whichtemporary contracts are used. Hence, I focus here on the following issues asbeing the ones with the main differences between the two countries, and discussthem in detail in the following subsections:

• Economic conditions, such as the segregation of the market regardingsectors (industry, agriculture, service), mobility of the labor force andmigration vs. immigration, labor market participation plus NEETs andthe duality of the labor market;

• Institutional frameworks such as minimum wages, Employment Protec-tion Legislation (EPL) and the education system.

5.1 Segregation of the market regarding sectors

The economic specialization of countries can influence the sensitivity of unem-ployment to cyclical conditions (Brada et al., 2014, Hutengs and Stadtmann,2014a). In Germany as well as in Poland, the trend in the service sector isclearly showing an increase in employment, while in the other sectors such asmanufacturing and agriculture the level of employment decreases, as shown inTable 9. In the service sector growth usually means that additional workforce isneeded and this would decrease the unemployment rate (Prybysz et al., 2000).But the share of workers between sectors as in Table 9 shows that the servicesector in Germany is higher than in Poland which could explain the higher levelin the unemployment rate in Poland.

8

5.2 Mobility of the labor force and migration vs. immigration

Regarding the mobility of the labor force, regional unemployment data fromEurostat (Eurostat, 2015) shows that immobility is still prevalent in both coun-tries. Between the regions in Poland as well as between the states in Germany,differences are still existent, for youth as well as for adults. Figures 4 and 5show the regional unemployment in Germany, respectively in Poland, in 2014.In Germany, there is still an East-West divide existing. Former East Germanyhas higher unemployment rates than former West Germany. In Poland there isneither an East-West nor a North-South divide, but the central region includ-ing Warsaw is having the lowest rates on both youth and adult unemployment.According to OECD (2014c), there are important impediments to internal labormobility, such as the quality of transport infrastructure and expensive urbanhousing, because of a lack of private rental supply.Mobility of labor forces does not only included the mobility within the country,but also international migration. According to OECD (2014d) the number ofPolish citizens who are staying abroad for more than three months increasedin 2012. Kaczmarczyk et al. (2014) shows already in 2011 an increase in of-ficially registered number of international emigrants. It was recorded that in2011 42.5% of migrating men and 49.2% of migrating women of the permanentemigrants were persons aged between 20 and 39 years. According to Kacz-marczyk et al. (2014) most of the Polish emigrants leave the country to workabroad. The most important migrant sending regions included in absoluteterms Slaskie, Malopolskie (both areas, so-called voivodeships, in region Polud-niowy), Dolnoslaskie (region Poludniowo-Zachodni) and Podkarpackie (regionWschodni) and in relative terms Podlaskie and Podkarpackie (both areas inregion Wschodni) (Kaczmarczyk et al., 2014). Regarding unemployment ratesthese are regions with the unemployment rates in the middle of the ranges,except for the region Wschodni which is the region with the highest unemploy-ment rates for youth as well as for adults (see Figure 5). Germany is one ofthe main destinations of Polish emigrants, which is shown in OECD (2014a).Poland is the top one country of origin of total inflows of foreigners in 2012as well as the annual average between 2002 and 2011. Germany does have apositive net immigration and according to OECD (2014a) this contributed toemployment growth. The number of younger foreign employees with tertiaryeducational qualification increased in 2011 as well as the employment rate offoreign workers aged 20 to 64 with a vocational background. The labor marketintegration of foreign workers has improved, the policy now focus on increasingthe employment rates of particular groups, such as women with a migration

9

background (OECD, 2014a). But on the other hand, also high educated Ger-man are leaving the country to work abroad (OECD, 2015c).

5.3 Labor market participation plus NEETs

As difficulties in finding work oblige some young persons to stay in school, tore-enter school and/or university, to start an apprenticeship etc., the labor par-ticipation rate of young persons should decrease. According to Dietrich (2012),changes in unemployment rate may be interpreted as an exchange between un-employed and employed (i.e. within labor force), but there can also be anexchange with an inactive group (outside the labor force). People in educationare not counted for the unemployment rate and for the labor force, so youthunemployment rate should decline. As can be seen in Figure 6, the labor mar-ket participation rate of youth (age cohort 15-24) in Poland and Germany isslightly decreasing, but there is no strong effect. According to Dietrich (2012),the decrease in youth labor market participation shows that a change in theyouth unemployment rate captures only part of the dynamic caused by thebusiness cycle and should be investigated further.But the labor participation rates do not include those young people that areoutside of the labor force. This group is called the „youth left behind“ and canbe proxied by the number of people who are neither employed nor in educationor training, so-called NEETs (Scarpetta et al., 2010). Figure 7 shows the pro-portion of youth who are not in employment and not in education or trainingfor Poland and Germany. The data is an indicator provided from the Inter-national Labour Organization (ILO, 2014), but only for the time period 2003(Germany) or 2004 (Poland) until 2013. For Germany, the share is decreasing,while for Poland it is increasing since 2008.

5.4 Duality of the labor market

According to Scarpetta et al. (2010), the dominant related factor for the higherbusiness-cycle sensitivity of youth is their high presence among those holdingtemporary jobs. With data from the Organisation for Economic Co-operationand Development (OECD, 2015b), the incidence of temporary employment foryouth and adults is shown for Germany in Figure 8 and for Poland in Figure9. Even though the incidence of temporary employment is increasing for adults(age cohort 25-54), there are large differences between youth and adults regard-ing temporary jobs. Young people in both countries have a larger proportionof temporary contracts. However, in Germany, temporary contracts are mainly

10

apprenticeship contracts (Scarpetta et al., 2010). Furthermore, it can be notedthat temporary contracts can be so-called „stepping stones“ to permanent con-tracts, i.e. the probability for youth of getting a permanent job after being ona temporary job is higher than after being unemployed (Scarpetta et al., 2010).As pointed out in OECD (2009), the high share of youth holding temporarycontracts in Poland can also indicate that there are structural rigidities on thelabor market which affect disproportionately youth, because it puts them at agreater risk to lose their job in economic recessions. They are the first to belaid off because the contracts are not extended or because they are subject tothe LIFO (last-in-first-out) rule. And strict employment protection on regularjobs could contribute to the high level of temporary contracts, because it re-strains employer’s willingness to take on a risk on workers without experience,i.e. new entrants in the labor market (OECD, 2009). Temporary contracts canalso be seen as dead-end jobs and a discussion can be found in Pastore (2015).But according to Baranowska et al. (2011), temporary contracts in Poland arerather used as a screening device for employers.

5.5 Minimum wages

As mentioned before, minimum wages are one labor market institution espe-cially relevant for youth unemployment (Brada et al., 2014). Germany intro-duced a national minimum wage in 2015, but already had minimum wages insome sectors determined by collective agreements before. Poland already hada national minimum wage in place covering all employees (ILO, 2015). Butas mentioned in OECD (2009), enterprises are allowed to pay new entrantsa reduced minimum wage during the first year of employment. Still e.g. La-poršek (2013) shows that the minimum wage tends to reduce youth employmentamong countries in the European Union with statutory minimum wage and thedisemployment effect is stronger among teenage workers. Also the differencesbetween countries, that set a lower minimum wage for young workers and theones that do not, the effect is the same. In Poland this might already explainpart of the unemployment rate for the new entrants in the market, also shownby OECD (2009).

5.6 Employment Protection Legislation (EPL)

Because the employment protection legislation and lay-off regulations affectmore the worker turnover and duration of unemployment than the unem-ployment level itself, they are more important for younger than for older

11

people (Brada et al., 2014). The OECD indicators on employment protectionlegislation in 2013 for Germany, Poland and the OECD unweighted averagefor comparative purpose are shown in Table 10 and include the protectionof permanent workers against individual and collective dismissals (EPRC),protection of permanent workers against (individual) dismissal (EPR), specificrequirements for collective dismissal (EPC) and regulation on temporary formsof employment (EPT) (OECD, 2015a). The scale is from 0 (least restrictions)to 6 (most restrictions). As can be seen in Table 10, Germany has stricterprotection of permanent workers and less stricter regulation on temporaryforms of employment than Poland (and the OECD unweighted average). Ifthe EPL is high on „permanent contracts“, than adults are in favor andthis can further increase the size and duration of unemployment for youth.The labor hoarding that took place in Germany with so-called „Kurzarbeit“(short-term work scheme) can be seen as such a practice (Choudhry et al., 2012).

5.7 Education system

According to Scarpetta et al. (2010) low-regulated labor markets provide asmoother school-to-work transition, but with highly regulated labor marketssuch as Germany it is very important to have strong vocational educationand training systems which can compensate the regulations. Germany hasestablished, besides the standard curricula, a professional system which allowsto combine work experience, on-the-job training and classroom teaching(Cahuc et al., 2013). According to Biavaschi et al. (2012) general schooling inGermany is followed by participation in upper secondary vocational educationas a standard pathway into the labor market. Vocational qualifications can beacquired by participating in one of the following options: (a) a dual vocationaltraining system with alternating school- and firm-based training, (b) full-timevocational schooling with a predominantly application-oriented curriculumor (c) tertiary education at vocational academies or universities. The dualapprenticeship system is generally seen as the main reason for the constantlylow youth unemployment rate in Germany, because it plays a central role withtwo thirds among the number of youths completing general schooling each yearwho enter the dual apprenticeship system, while about one fifth participate infull-time vocational schooling (Biavaschi et al., 2012). But, albeit its success,this system is not very easy to implement in other countries, because itrequires a big effort by all the partners involved, such as the social partners,public employment service (PES), and educational institutions (Pastore, 2015,

12

Biavaschi et al., 2012). And Scarpetta et al. (2010) found that in economicdownturns employers become reluctant to offer apprenticeships, especiallyto those youth lacking educational qualifications and from an immigrantbackground.In Poland, there exists three secondary schooling tracks which include generalupper-secondary schools (so-called lyceum), technical upper-secondary schools(so-called technikum) and basic vocational schools. Graduates of generalupper-secondary schools can continue their education in a postsecondaryschool and can receive a vocational diploma confirming vocational qualifica-tions in a given occupation (EURYDICE, 2014). According to Baranowskaet al. (2011) time spent in basic vocational schools is shorter, but they preparestudents mainly for manual occupations, while technikums have a longerduration, but provide a mix of general and vocational education for thepreparation for skilled service and technical occupations and offer the studentsalso the possibility to transfer into tertiary education. General secondaryschools provide no occupational qualifications, but prepare the students forhigher education. Firm-based training is emphasized more in the curriculumof basic vocational schools than in the technikum, but employer involvementin the design and organization of training decreased in the course of economicrestructuring (Baranowska et al., 2011). Higher education, i.e. tertiaryeducation, includes degree programs, provided by public and non-publicuniversity-type and non-university higher education institutions, and furthercollege programs, provided by colleges of social work, teacher training collegesand foreign-language teacher training colleges, where the latter two types arephased out now (EURYDICE, 2014). OECD (2009) explains as well that largesegments of firm-based vocational education already collapsed with the state-owned firms in the economic restructuring, so it is now more school-based. AndPolakowski (2012) confirms that the cooperation of schools and companies arelow. But Baranowska et al. (2011) show that graduates of secondary vocationalschools have faster transitions to employment than general secondary schoolgraduates, especially with vocational education and firm-based training, al-though that does not include better chances to transfer to open-ended contracts.

13

6 Conclusions and recommendations

In this paper I examined the development of the youth unemployment rate inGermany and Poland, using the estimates of age-cohort specific Okun coeffi-cients. The main empirical results can be summarized as follows:

1. Germany: The Okun coefficient for young people is larger than for otherage cohorts in absolute value, so youth are more sensitive to the businesscycle than adults, but the differences between the age cohorts are smalland not statistically significant.

2. Poland: The Okun coefficient for young people is larger than for other agecohorts in absolute value, so youth are more exposed to fluctuations thanother age cohorts. The differences between the age cohorts, especiallybetween 15-24 years old and the subsequent age cohort of the 25-34 yearsold, are large, but not statistically significant.

3. The Okun coefficient for young people differs between the two countries,showing that young Polish people are hit harder by macroeconomic shocksin comparison to young German people. This result not only holds foryouth, but for all age cohorts.

Any policy recommendation here should consider GDP growth, because youthunemployment is more sensitive to business fluctuations and it is a relevant fac-tor for adult unemployment as well. And without economic growth no youthpolicy can ever be effective (Pastore, 2015). But according to Polakowski (2012),the Polish economy has grown comparatively fast, but without creating newjobs. There had been a decrease in employment in agriculture and the growthof the service sector, but also a decrease in demand for labor due to increasedlabor productivity. Still, further promotion of the service sector in Poland aswell as in Germany is recommended as this could lead to a higher growth ofthe job market (Prybysz et al., 2000, OECD, 2014b).The obstacles to internal mobility in Poland should be reduced e.g. by continu-ing to develop transport infrastructure, such as the quality of the rail network,and reforming housing policies (OECD, 2014c).The number of NEETs in Poland are rising which is a reason of concern. Scar-petta et al. (2010) proposes a better cooperation between employment servicesand education system to reach youth as soon as there is a risk of disengage-ment, as well as an early guidance to school-leavers in search of a job and a„learn/train-first“ approach to maintain youth connected to the labor market.

14

If temporary contracts in Poland are used as screening device, as discoveredby Baranowska et al. (2011), then qualifications such as certificates are notsignaling the quality of the certificate holders to the companies. Therefore,as suggested by OECD (2009), a universal Vocational Education and Training(VET) classification system should be implemented in Poland. Further propos-als are made by OECD (2014c), such as enhancing work-based learning in VETprograms by boosting social partners’ involvement and raising the quality ofteaching as well as strengthening the link with businesses.Scarpetta et al. (2010) proposes to rebalance the employment protection, sothat youth can gradually move from entry jobs to career employment, i.e. asmooth transition from temporary to more stable and rewarding jobs whichcould reduce the labor-market duality and the sensitivity of youth to businesscycles. OECD (2009), too, suggests to reduce the gap in employment protec-tion between open-ended, fixed-term contracts and the „commission contracts“in Poland and OECD (2014b) recommends to reduce the gap in employmentprotection between permanent and temporary workers in Germany.The major challenge for Germany is the labor market integration of young peo-ple failing to enter regular vocational training (Biavaschi et al., 2012). Appren-ticeships for unskilled young people and support measures to help apprenticeswhose contracts had been ended to complete their training should be included(Scarpetta et al., 2010).All in all, my proposals are:

• in the short-term: job-search assistance and guidance for all youth bypublic employment services (Scarpetta et al., 2010);

• in the long-term: for Poland structural reforms regarding the educationsystem, employment protection and mobility as described above; for Ger-many strategies to avoid school drop-outs and offerings of a second chanceof qualification for every young person (Scarpetta et al., 2010).

Acknowledgement

The research underlying this paper was supported by the Deutsch-PolnischeWissenschaftsstiftung. A preliminary version of this paper was presented at the„1st Workshop (Youth) Unemployment in Europe“ at the European UniversityViadrina, April 16-17, 2015, and I am grateful for comments by the workshopparticipants.

15

ReferencesBaranowska, A., Gebel, M., and Kotowska, I.E. (2011), The role of fixed-termcontracts at labour market entry in Poland: stepping stones, screening de-vices, traps or search subsidies?, Work, employment and society, 25(4), pp.777–793.

Bell, D. and Blanchflower, D. (2011), Young People and the Great Recession,Oxford Review of Economic Policy, 27 (2), pp. 241–267.

Berlingieri, F., Bonin, H., and Sprietsma, M. (2014), Youth Unemploymentin Europe - Appraisal and Policy Options, Centre for European EconomicResearch / Zentrum für Europäische Wirtschaftsforschung GmbH (ZEW).

Biavaschi, C., Eichhorst, W., Giulietti, C., Kendzia, M.J., Muravyev, A.,Pieters, J., Rodríguez-Planas, N., Schmidl, R., and Zimmermann, K.F.(2012), Youth Unemployment and Vocational Training, IZA Discussion Pa-per, 6890.

Boulhol, H. and Sicari, P. (2013), Labour Market Performance by Age Groups:A Focus on France, OECD Economics Department Working Papers, 1027.

Brada, J.C., Marelli, E., and Signorelli, M. (2014), Young People and the La-bor Market: Key Determinants and New Evidence, Comparative EconomicStudies, 56, pp. 556–566.

Cahuc, P., Carcillo, S., Rinne, U., and Zimmermann, K.F. (2013), Youth Un-employment in Old Europe: The Polar Cases of France and Germany, IZADiscussion Paper, 7490.

Choudhry, M.T., Marelli, E., and Signorelli, M. (2012), Youth Unemploymentand the Impact of Financial Crises, International Journal of Manpower,33(1), pp. 76 – 95.

Dietrich, H. (2012), Youth Unemployment in Europe - Theoretical Con-siderations and Empirical Findings, Friedrich Ebert-Stiftung Interna-tional Policy Analysis, manuscript available at: http://library.fes.de/pdf-files/id/ipa/09227.pdf.

EC (2015), Domestic products statistics, European Commission, avail-able at: http://ec.europa.eu/economy_finance/ameco/user; last access on:2015/08/03.

Eurostat (2015), Employment and unemployment (LFS), European Commis-sion, available at: http://ec.europa.eu/eurostat/web/lfs/data/main-tables,last access on: 2015/08/04.

EURYDICE (2014), The System of Education in Poland, Foundation for theDevelopment of the Education System (FRSE), Polish EURYDICE Unit.

i

Hutengs, O. and Stadtmann, G. (2014a), Age- and Gender-Specific Unemploy-ment in Scandinavian Countries: An Analysis based on Okun’s Law, Com-parative Economic Studies, 56(4), pp. 567–580.

Hutengs, O. and Stadtmann, G. (2014b), Don’t trust anybody over 30: youthunemployment and Okun’s law in CEE countries, Bank i Kredyt, 45(1), pp.1–16.

ILO (2014), ILOSTAT Database, International Labour Organization, avail-able at: http://www.ilo.org/ilostat/faces/home/statisticaldata; last accesson: 2014/12/18.

ILO (2015), ILO Working Conditions Laws Database, ILO, Geneva, availableat: http://www.ilo.org/dyn/travail; last access on: 2015/08/14.

Kaczmarczyk, P., Da̧browski, P., Fihel, A., and Stefanska, R. (2014), RecentTrends in International Migration in Poland: The 2012 SOPEMI Report,CMR Working Papers, 71/129.

Knotek, E. (2007), How Useful is Okun’s Law?, Economic Review, FederalReserve Bank of Kansas City, fourth quarter, pp. 73–103.

Laporšek, S. (2013), Minimum wage effects on youth employment in the Euro-pean Union, Applied Economics Letters, 20(14), pp. 1288–1292.

Mroz, T. and Savage, T. (2006), The long-term effects of youth unemployment,The Journal of Human Resources, 41(2), pp. 259–293.

OECD (2009), Jobs for Youth/Des emplois pour les jeunes: Poland, OECDPublishing.

OECD (2014a), Germany, International Migration Outlook 2014, OECD Pub-lishing.

OECD (2014b), OECD Economic Surveys: GERMANY, OECD Publishing.

OECD (2014c), OECD Economic Surveys: POLAND, OECD Publishing.

OECD (2014d), Poland, International Migration Outlook 2014, OECD Publish-ing.

OECD (2015a), Employment Protection Database, Organisationfor Economic Co-operation and Development, available at:http://www.oecd.org/employment/emp/oecdindicatorsofemploymentprotection.htm; last access on: 2015/03/03.

OECD (2015b), Labour force statistics, Organisation for Economic Co-operationand Development, available at: https://data.oecd.org/; last access on:2015/07/29.

OECD (2015c), Talent Abroad: A Review of German Emigrants, OECD Pub-lishing.

ii

O’Higgins, N. (1997), The challenge of youth unemployment, International So-cial Security Review, 50(4), pp. 63–93.

O’Higgins, N. (2003), Trends in the Youth Labour Market in Developing andTransition Countries, Social Protection Discussion Paper, 321.

O’Higgins, N. (2012), This Time It’s Different? Youth Labour Markets DuringThe Great Recession, Comparative Economic Studies, 54, pp. 395–412.

Okun, A.M. (1962), Potential GNP: its measurement and significance, M.N.Baily and A.M. Okun (eds.), The battle against unemployment and inflation,Norton, New York.

Pastore, F. (2015), The Youth Experience Gap - Explaining National Differencesin the School-to-Work Transition, Springer, Cham.

Polakowski, M. (2012), Youth Unemployment in Poland, FriedrichEbert-Stiftung International Policy Analysis, manuscript available at:http://library.fes.de/pdf-files/id/09477.pdf.

Prybysz, K., Dziechciarz, J., and Siedlecki, J. (2000), Statistical Analysis ofYouth Unemployment in Poland, Poverty and Unemployment, pp. 129–142.

Scarpetta, S., Sonnet, A., and Manfredi, T. (2010), Rising Youth Unemploy-ment During the Crisis: How To Prevent Negative Long-term ConsequencesOn A Generation?, OECD Social, Employment and Migration Papers, 106.

Unt, M. (2012), Boom and bust effects on youth unemployment in Estonia,Friedrich Ebert-Stiftung International Policy Analysis, manuscript availableat: http://library.fes.de/pdf-files/id/09473.pdf.

iii

Appendix

Regression analysis, tables and graphics

Table 1: Youth unemployment vs. adult unemployment (O’Higgins, 2012).Country coefficient R2 N

Germany 1.0070*** 0.5436 25(0.1924)

Poland 2.1856*** 0.9472 23(0.1126)

EU-15 2.1765*** 0.8974 25(0.1535)

Source: Own elaboration with data from OECD (2015b). Notes: N - number of observations; standard errors inparentheses; significance at *** 1% level, ** 5% level, * 10% level.

Table 2: Results Breusch-Pagan-Test for heteroscedasticity.Country BP p-Value

Germany 17.2055 0.0456Poland 26.1937 0.0019EU-15 19.1254 0.0242

Nullhypothesis: Homoskedasticity.

Table 3: Results Durbin-Watson-Test for serial correlation.Country DW p-Value

Germany 1.0771 0.0000Poland 1.0428 0.0000EU-15 1.4533 0.0002

Nullhypothesis: No autocorrelation.

Table 4: Panel Regression Results with fitted MA(1) residuals.Country 15 − 24 25 − 34 35 − 44 45 − 54 55 − 64 R2 N

Germany −0.3258*** −0.2875** −0.1789* −0.1728 −0.1411 0.3548 115(0.0895) (0.0893) (0.0891) (0.0891) (0.0898)

Poland −1.1360*** −0.6093* −0.4972* −0.4144 −0.2601 0.4481 110(0.2378) (0.2362) (0.2359) (0.2361) (0.2359)

EU-15 −0.7252*** −0.4600*** −0.3077*** −0.2572*** −0.2393*** 0.7067 115(0.0624) (0.0621) (0.0620) (0.0620) (0.0625)

Source: Own elaboration with data from OECD (2015b). Notes: N - number of observations; standard errors inparentheses; significance at *** 1% level, ** 5% level, * 10% level.

iv

Table 5: Wald test for equality of coefficients - Germanyβ25−34 β35−44 β45−54 β55−64

β15−24 0.0922 1.3545 1.4693 2.1415β25−34 0.7423 0.8279 1.3477β35−44 0.0023 0.0895β45−54 0.063

Notes: significance at *** 1% level, ** 5% level, * 10% level.

Table 6: Wald test for equality of coefficients - Polandβ25−34 β35−44 β45−54 β55−64

β15−24 2.486 3.6506 4.6633* 6.8601*β25−34 0.113 0.3416 1.096β35−44 0.0616 0.5055β45−54 0.2141

Notes: significance at *** 1% level, ** 5% level, * 10% level.

Table 7: Wald test for equality of coefficients - EU-15β25−34 β35−44 β45−54 β55−64

β15−24 9.1537** 22.653*** 28.482*** 30.737***β25−34 3.023 5.3589* 6.3322*β35−44 0.3318 0.6063β45−54 0.0416

Notes: significance at *** 1% level, ** 5% level, * 10% level.

Table 8: Unemployment rates and labor market participation rates in % in2014.

Country age cohort 15-24 age cohort 25-34

Unemployment Labor Market Unemployment Labor Marketrate Participation rate rate Participation rate

Germany 7.76 49.95 5.79 84.95Poland 23.87 33.86 9.79 85.64EU-15 21.64 45.70 12.49 84.68

Source: Own elaboration with data from OECD (2015b).

v

Table 9: Employment changes between 2007 and 2014 and share of workers insectors in % in 2014.Country Agriculture Construction Industry Manufacturing Service

2007- share 2007- share 2007- share 2007- share 2007- share2014 2014 2014 2014 2014

Germany −33.63 1.19 8.20 5.73 −4.62 17.73 −6.92 16.38 9.11 58.97Poland −18.74 9.63 12.59 6.28 0.61 19.31 −3.98 16.07 10.65 48.72

Source: Own elaboration with data from OECD (2015b). Industry is excluding Construction.

Table 10: OECD indicators on EPL in 2013.Country EPRC EPR EPC EPT

Germany 2.98 2.72 3.63 1.75Poland 2.39 2.20 2.88 2.33OECD unweighted average 2.29 2.04 2.91 2.08

Source: Own elaboration with data from OECD (2015a).

Figure 1: GDP Growth.

Source: Own elaboration with data from EC (2015).

vi

Figure 2: Youth unemployment rate (age cohort 15-24).

Source: Own elaboration with data from OECD (2015b).

Figure 3: Youth-adult unemployment rate ratio.

Source: Own elaboration with data from OECD (2015b).

Figure 4: Incidence of regional unem-ployment in Germany in 2014.

Source: Own elaboration with data from OECD (2015b).

Figure 5: Incidence of regional unem-ployment in Poland in 2014.

Source: Own elaboration with data from OECD (2015b).

vii

Figure 6: Labor market participation rate youth (age cohort 15-24).

Source: Own elaboration with data from OECD (2015b).

Figure 7: Youth (age cohort 15-24) not in employment and not in education ortraining (NEET).

Source: Own elaboration with data from ILO (2014).

Figure 8: Incidence of temporary em-ployment in Germany.

Source: Own elaboration with data from OECD (2015b).

Figure 9: Incidence of temporary em-ployment in Poland.

Source: Own elaboration with data from OECD (2015b).

viii

Deckblatt_Discussion paper_Vorlage.pdfDunsch_Okun'slaw.pdf